The allocation of ministerial posts (portfolios) is the main outcome of the government formation process, and central to what “to govern” means (Laver and Shepsle, Reference Laver and Shepsle1996; Martínez-Gallardo, Reference Martínez-Gallardo2010; Golder, Reference Golder2015). Portfolios provide office benefits (e.g., patronage, perks, and control of budgets) to the party that holds them and also influence the overall direction of government policy, who formulates it, and the way the policy will be implemented. These payoffs represent the bottom line of the executive political process in coalition governments, and the understanding of their distribution can elucidate who gets to govern, and how governments formulate and enact their policies.
In parliamentary systems, there is evidence that among the parties that enter into a government together, cabinet portfolios are roughly allocated in proportion to the number of seats each party brings to the coalition (Schofield and Laver, Reference Schofield and Laver1985; Laver and Schofield, Reference Laver and Schofield1990; Laver et al., Reference Laver, de Marchi and Mutlu2011; Falcó-Gimeno and Indridason, Reference Falcó-Gimeno and Indridason2013; Bergman et al., Reference Bergman, Ersson and Hellström2015).Footnote 1 Given the absence of the vote of no confidence in presidential systems, formateur parties in presidential democracies are not reliant on their coalition partners for their continued survival in office. As a result, scholars suggest that the distribution of portfolios in presidential systems should occur in a disproportional fashion (Amorim Neto, Reference Amorim Neto2006a,Reference Amorim Netob; Ariotti and Golder, Reference Ariotti and Golder2018). The expectation of a higher share of portfolios for the president's party has been supported in the literature and defined as the formateur's advantage or the formateur's bonus (Altman, Reference Altman2000; Amorim Neto and Samuels, Reference Amorim Neto and Samuels2010; Ariotti and Golder, Reference Ariotti and Golder2018; Chaisty et al., Reference Chaisty, Cheeseman and Power2018).
In this study, I show that while the president's party does typically receive more portfolios than predicted by their seat share contribution on average, there is significant variation in the size of the portfolio share held by the formateur's party in presidential systems. I then develop a theory to explain when a greater formateur's advantage can be expected within presidential democracies, providing a more-nuanced comparison between the dependence of the executive upon the legislature in parliamentary and presidential forms of government.
Because they depend on a legislative majority to survive and to legislate, prime ministers are more dependent upon the legislature than presidents. However, the absence of a no-confidence constraint in presidential systems does not make presidents less dependent on the legislature to make and enact policies. The need for all executives to make policies means that all presidents are dependent upon the legislature to some degree. The degree of the ability of the president to shape the legislative agenda has clear implications on how governments are formed, and when presidential cabinets will look similar to parliamentary cabinets. Presidents form coalitions—i.e., allowing other parties to control portfolios—when they need support from other parties to accomplish their legislative goals. Therefore, the greater the power of the presidents to influence the policy agenda in the legislature, the less they will relinquish control of portfolios. As a result, there will be a correlation between president's policy-making powers and the size of the formateur's advantage. The empirical implication is that when presidents have greater institutionally-granted policy-making powers, a greater formateur's advantage should be observed.
The theoretical framework here advanced is based on an existing general argument about the distribution of portfolios by the president in exchange for agenda-setting power in the legislative process (for instance, see Shugart and Carey, Reference Shugart and Carey1992; Mainwaring and Shugart, Reference Mainwaring and Shugart1997; Amorim Neto, Reference Amorim Neto2006b; Amorim Neto and Samuels, Reference Amorim Neto and Samuels2010; Alemán and Tsebelis, Reference Alemán and Tsebelis2011; Kellam, Reference Kellam2015). Yet, in this study, this argument is tested in a new way, identifying, for the first time in the study of coalition governments in presidential democracies, the direction of the proportionality deviation—i.e., whether to the president's party benefit or not.
As supported by the results from new data on government formation in 20 presidential systems over 73 years (1946–2019), if the president is more dependent upon the legislature to make and enact policies, the balance of power in presidential cabinets is more likely to reflect the balance of power in the legislature. In this case, the presidential cabinet can be very similar to the proportional cabinets usually formed in parliamentary systems. However, when institutional provisions allow some presidents greater leeway in making and enacting policies, then a more disproportional distribution of portfolios to the benefit of the president's party—a formateur's advantage—is expected. The findings of this study enhance our understanding on when and why we should expect similarities and differences in coalition dynamics between presidential and parliamentary democracies, and in particular the dynamics of coalition formation within presidential systems.
1 Portfolio allocation in parliamentary and presidential democracies
It is frequently the case in multiparty systems that no single party holds an absolute majority of seats in the national legislature. This circumstance usually compels heads of government to form a coalition in order to govern effectively. This inevitably involves horse trading between elites to determine how the government will be formed and which portfolios will be allocated to those parties that comprise the coalition.Footnote 2
Heads of government distribute ministerial portfolios across parties in exchange for legislative support. Legislative parties value portfolios, either because of their intrinsic benefits (e.g., access to office perks and patronage) or because portfolios provide the opportunity to shape the policy agenda of the government.
In parliamentary systems, the formateur of the coalition government most often comes from the largest party in the legislature and generally becomes the new head of government (e.g., prime minister) if the formation succeeds. Because the prime minister depends on legislative parties for her government's survival, the prime minister cannot afford to ignore the preferences of the parties that will comprise the government (Amorim Neto and Samuels, Reference Amorim Neto and Samuels2010). Ignoring these preferences could result in a legislative majority removing the government from office by passing a vote of no confidence. If this happens, neither the formateur's party nor the parties that comprise the new legislative majority can be confident of a greater seat share in the next government formation attempt. Therefore, the presence of the vote of no confidence in parliamentary systems deters the formateur from forming a cabinet too disproportional in her party's favor (Golder and Thomas, Reference Golder and Thomas2014).
From the absence of the vote of no confidence in presidential systems, formateur parties are not reliant on their coalition partners for their continued survival in office. As a result, scholars suggest that formateurs in presidential systems will value the contribution of the legislative seats of their coalition partners to the government less than would be the case in parliamentary systems (Ariotti and Golder, Reference Ariotti and Golder2018). This means that non-formateur parties cannot expect the offers they receive to be as generous from presidential formateurs as from prime ministerial formateurs. It follows that formateur parties should receive a higher share of portfolios, relative to their legislative size, in presidential systems than in parliamentary systems (Amorim Neto and Samuels, Reference Amorim Neto and Samuels2010; Indridason, Reference Indridason2015). As a consequence, while we should expect a proportional distribution of portfolios in parliamentary systems, in presidential systems we should observe a disproportional distribution to the benefit of the formateur's party. The literature already reveals a bonus in the portfolio share for the president's party on average (Altman, Reference Altman2000; Amorim Neto and Samuels, Reference Amorim Neto and Samuels2010; Indridason, Reference Indridason2015; Ariotti and Golder, Reference Ariotti and Golder2018). Yet, as I demonstrate below, there is a substantial variation in the allocation of portfolios to presidential parties to be explained.
Figure 1 depicts the variation in the allocation of portfolios in parliamentary and presidential democracies, identifying the formateur's party. Each dot in Figure 1 represents a party that ended up in the coalition government, with solid dots identifying the formateur's party. Observations located on the 45-degree dashed line (i.e., a perfect proportionality or Gamson's Law) are parties that received a proportional share of portfolios based on their seat share contribution to the coalition. Observations above the dashed line are parties that received more portfolios than expected based on the number of legislative seats they contribute to the coalition, while parties below the dashed line are parties that received less portfolios than expected based on their number of legislative seats.Footnote 3
Figure 1. Proportionality in portfolio allocation in parliamentary and presidential systems, (a) parliamentary systems, (b) presidential systems.
Comparing the plots of Figure 1, it is noteworthy that seat share contribution accounts for much more of the variation of portfolio share in parliamentary systems than in presidential systems. The higher variation of the solid black dots identifying the formateur's party above and below the Gamson's Law line in Figure 1b suggests that something else, besides seat share contribution, could explain the variance in portfolio share in presidential democracies.
Building on existing arguments about the effect of presidential institutional powers on the formation and composition of executive cabinets (Shugart and Carey, Reference Shugart and Carey1992; Mainwaring and Shugart, Reference Mainwaring and Shugart1997; Amorim Neto, Reference Amorim Neto2006b; Amorim Neto and Samuels, Reference Amorim Neto and Samuels2010; Alemán and Tsebelis, Reference Alemán and Tsebelis2011; Kellam, Reference Kellam2015), in the next section I theorize that the extent to which the president controls the legislative process is a crucial factor to understanding the variation in the allocation of portfolios to presidential parties. I then demonstrate it identifying the direction of the proportionality deviation, for example, whether to the president's party benefit or to junior partners of the coalition.
2 Explaining the formateur's advantage within presidential democracies
In both parliamentary and presidential forms of government, chief executives seek to endure in office (and to reap the political benefits that come from controlling ministries), and to legislate (i.e., to influence the policy agenda). In parliamentary systems, coalition formation by the formateur has a direct effect on survival, through the no-confidence procedure (no-confidence constraint), and a direct effect on policy through coalition parties’ commitment to vote for the agenda of the government (legislative constraint). In presidential systems, however, coalition formation can affect policy but not the survival of the government (at least not directly).
While in parliamentary systems, the government usually exerts control over legislation, in presidential systems, the policy-making powers of the president can vary significantly (Shugart and Carey, Reference Shugart and Carey1992; Alemán and Schwartz, Reference Alemán and Schwartz2006; Cheibub, Reference Cheibub2007). Presidents have a greater or lesser influence on policy, conditional on how dependent they are on the legislature to make and enact policies. This dependence relies on presidents’ policy-making powers granted by constitutions.Footnote 4 The bargaining process between the president and the legislature around a coalition formation, therefore, depends on institutional arrangements that define the balance of power between the president and legislative parties and their abilities to influence the policy agenda. The greater the president's policy-making powers, the less dependent the president is on the legislature to legislate (the smaller the legislative constraint), and the greater the president's ability to shape the legislative agenda. Consequently, this leads to a greater presidential control over resources and benefits of office, allowing presidents to control a higher share of portfolios than the share of legislative seats her party contributes to the coalition. Otherwise, the greater the legislative constraint experienced by the president, more dependent is the president upon the legislature to legislate and have her policies approved. Under this context, we should see a more proportional distribution of portfolios (as we commonly see in parliamentary systems), i.e., a distribution of power within the executive cabinet that resembles the distribution of power within the legislature.
According to the theory here advanced, a greater share of portfolios controlled by presidential parties should be observed in countries where, ceteris paribus, presidents have a greater opportunity to influence the policy agenda in the legislature. The greater the policy-making powers of the presidents, the less dependent they are on the legislature to legislate, and the less they will relinquish control of portfolios. As a result, the empirical implication to be tested in this study is: the greater the president's policy-making powers, the greater the president's party portfolio share.
3 Empirical strategy
The unit of analysis in this study is the coalition partyFootnote 5 annually considered. The empirical analyses were conducted using new data from 20 presidential multiparty democracies worldwide.Footnote 6 The data comprise a total of 117 unique coalition parties (among those, 48 unique presidential parties), spanning over 73 years (1946–2019).Footnote 7
The dependent variable—portfolio share—captures the distribution of cabinet portfolios among the parties that comprise the coalition government. Thus, this variable indicates the percentage of portfolios coalition parties (i.e., either the formateur or junior parties) receive from the total number of portfolios available.Footnote 8
The main independent variable in the analysis—presidential power—captures the institutional strength of the executive authority, as developed by Doyle and Elgie (Reference Doyle and Elgie2014). The literature is prolific on measuring presidential power, and there are well-known problems with each of these measurements (Fortin, Reference Fortin2012). To maximize the reliability of presidential power measurement, Doyle and Elgie (Reference Doyle and Elgie2014) generated a time-series cross-sectional dataset of presidential power scores with country years as the unit of observation, based on 28 existing measures in the literature. This is the most comprehensive measurement of presidential power (both in terms of countries and years covered), and using a single index leveraging the reliability of other existing indices help us avoid empirical results that would be too sensitive for the particular measure that is used. The presidential power scores range from 0 to 1 in separate time periods following constitutional changes of a country's presidential powers. Greater values of the index indicate that the presidency is a powerful actor within the decision-making and legislative-making processes of the country.Footnote 9 Based on the empirical implication to be tested in this study, a greater formateur's advantage is expected as presidential power increases.
To capture the effect of presidential power on the portfolio share held by the formateur, an interactive term between formateur and presidential power should be added to the model specification. The formateur's party is identified by a dichotomous variable—formateur—assuming the value of 1 for the formateur's (or president's) party, and the value of 0 otherwise. The variable seat share contribution indicates the percentage of legislative seats coalition parties contribute to the total number of legislative seats held by the government's coalition, when the cabinet is appointed by the formateur.
Two control variables are included in the regression models to account for possible confounders and to isolate the effect of presidential power. First, a dummy for electoral year is added based on the empirical expectation that as the next election approaches, parties have fewer incentives to join or remain in the government, and therefore cabinet termination should be more likely (Chasquetti, Reference Chasquetti1999; Altman, Reference Altman2000; Alemán and Tsebelis, Reference Alemán and Tsebelis2011). Second, presidential majority indicates whether the president's party alone holds the majority of seats in the legislature. Presidential minority situations in the legislature immediately after the elections exert a strong pressure on the president to form a coalition government (Cheibub, Reference Cheibub2007).
4 Empirical analysis: results and interpretation
Figure 2 presents the results for the main empirical implication tested in this study, i.e., the greater the presidential power, the greater the portfolio share of the president's party. Controlling for electoral year and presidential majority, the results support this empirical expectation. As depicted in Figure 2a, the estimate for the interactive term between presidential power and formateur is positive and statistically significant at level 0.05, indicating that the greater the institutional powers of the president, the greater the formateur's portfolio share. In substantive terms, on average and holding the control variables constant, a country with a strong president (in terms of institutional executive prerogatives) leads to a bonus of approximately 23 percent of the portfolios available to the benefit of the president's party. Seat share contribution is still a strong predictor for portfolio share, and it is far from a one-to-one proportionality.
Figure 2. The effect of presidential powers on the formateur's portfolio share. (a) Estimates and confidence intervals. (b) Marginal effects. Notes: Dependent variable: portfolio share. Model specified with country and year fixed-effects. 95% confidence level. A full regression report can be viewed in Model 3 of Table E.1 in the Supplementary Material.
We can also highlight the main result of this study, considering the changes in the effect of formateur on portfolio share conditional on the variation of the presidential power. As we can see in Figure 2b depicting this marginal effect, the effect of formateur on portfolio share increases as the presidential power increases. Therefore, the stronger the institutional powers of the president, the greater the effect of formateur on portfolio share. Considering the range of presidential power in the sample (from 0.03 [the weakest presidency] to 0.61 [the strongest presidency]), the president's party receives a bonus of approximately 10 percent of the portfolios available when we move from the weakest presidency to the strongest presidency; from a bonus of 10 percent (from being the formateur's party with weak institutional powers) to a bonus of approximately 23 percent (from being the formateur's party with strong institutional powers).
4.1 Robustness checks
The results provide support for the theory advanced in this study. On average, and controlling for possible confounders, a greater bonus of portfolio share to the president's party is found in those countries in which the president is less dependent on the legislature, and has a greater opportunity to formulate, influence, and enact her policy agenda, i.e., presidents that are constitutionally empowered with greater policy-making powers. To further evaluate the consistency of this finding, I conduct several robustness checks, including the impact of salient portfolios on this result.Footnote 10
To identify the effect of extreme values of my measurement of presidential power on the estimates of my model, a robustness test removing one country at a time in the analysis is presented in Appendix G of the Supplementary Material. Although in the expected positive direction, the estimate for the interactive term between formateur and presidential power is not statistically significant at conventional levels when Indonesia (the case with the weakest presidency in my sample, with a score of 0.03 in presidential power) is removed.Footnote 11
The estimates for the models weighting for the salience of the ministry of finance—considered the most-important ministerial post to be allocated by presidents in presidential democracies (Martínez-Gallardo, Reference Martínez-Gallardo2010; Batista, Reference Batista2017)—suggest that strong presidents might have an advantage in the allocation of portfolios that is not only quantitative, but also qualitative (i.e., controlling the most-important portfolios). However, as indicated by the overlapping confidence intervals (at level 0.05), this result is far from conclusive and worthy of exploration in future research.Footnote 12
5 Conclusion
A stable finding in the studies of coalition formation in presidential systems asserts that the absence of a dissolution constraint in presidential democracies leads to the expectation of a formateur's advantage in the allocation of portfolios. I provide empirical evidence of a substantial variation in the allocation of portfolios to presidential parties, and advance a theory for this variation that is consistent with this expectation but also makes novel predictions.
While the president's party receives, on average, more portfolios than predicted by their seat share contribution, in this study I demonstrate with theory and data that the variation we see in the formateur's advantage can be explained by the trade-off between the president's desire to control as many portfolios as possible, and her desire to legislate and influence policy. When the president is less dependent on the legislature to formulate and enact her policies, she relinquishes less control of portfolios to other members of the coalition. As a consequence, presidents with greater institutionally-granted powers to influence the policy agenda in the legislature hold, on average, more portfolios than the share of legislative seats their parties contribute to the coalition. Still, the balance of power in presidential cabinets can reflect the balance of power in the legislature when the head of government is more dependent on the legislature to legislate. In this case, the presidential cabinet can be very similar to the proportional cabinets usually formed in parliamentary systems.
By conceiving the dependence of the president on the legislature as a continuous variable that varies with the president's ability to make and enact policy in the legislature, this study enhances our understanding of key aspects of executive-legislative relations in different systems of government, and, particularly, when and why we should expect similarities across presidential and parliamentary democracies with regards to coalition government formation. This study also highlights that while there are important differences between presidential and parliamentary systems, a unified framework can lead to a prolific research agenda.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/psrm.2022.20.
Acknowledgments
The author would like to thank Alejandro Medina, Guy Whitten, José Cheibub, Kirby Goidel, Marcelo Camerlo, Maria Escobar-Lemmon, Michelle Taylor-Robinson, Samuel Müller, Thiago Talzzia, Thomas Bräuninger, William Clark, three anonymous reviewers, and the editors of Political Science Research and Methods for their invaluable comments. The author acknowledges funding support from the German Research Foundation (grant number 139943784) via Collaborative Research Center (SFB) 884, “The Political Economy of Reforms” at the University of Mannheim.
Data availability statement
The replication file and the data that support the findings of this study are openly available in Harvard Dataverse at https://doi.org/10.7910/DVN/RHMSKP.
Open access
